Testing Qwen 3.8 Flash Next UD-Q4_K_XL on four RTX 3090s showed it underperformed compared to Qwen 3.8 27b Q8, with slower Strata results and no playable executable after a four‑hour run. Subsequent runs were limited to one hour, and initial prompt bugs in the 27b model were self‑corrected.
Read original
reddit/r/LocalLLaMA